Toward a detailed model of processing for language describing the physical world

摘要

This paper explores the problem of judging whether or not an English sentence could correspond to a real world situation or event which is literally, physically plausible, and the related problem of representing the different possible physical situations. The judgement of plausibility can be made at a high level by checking semantic marker restrictions on verb case frame constituents. Often, however, plausibility judgement can only be based on the results of an attempt to construct (imagine) a scene that corresponds to the sentence, and which does not violate sense (i.e. relevant physical laws and expected, stereotyped behavior). Methods are presented for constructing representations for different scenes which could correspond to a sentence. These methods incorporate (1) subscripts (sequences of scenes which comprise an event, with attached preconditions and postconditions) to express different verb senses, (2) object representations which express properties such as shape, size, weight, strength, and behavior under common conditions; (5) physical laws, encoded as constraints on behavior; (4) representation of context; and (5) robot problem solving-like methods to fit all this material together.